According to a study published in Arthritis Care & Research, diagnostic prediction modeling for knee osteoarthritis (OA) in adolescent men helps calculate the individual and population risk of developing knee osteoarthritis over 40 years. In addition, it encourages preventive and lifestyle modifications to reduce long-term risk for knee OA.
The study cohort included 40,118 young men who underwent mandatory military conscription in Sweden from 1969 to 1970 and were 18 years of age; baseline data were linked to diagnostic codes for OA drawn from the Swedish National Patient Register between 1987 and 2010. Following the Nottingham model for predicting OA, the original study sample from which the model was developed included predictors of gender, age, body mass index (BMI), knee injury, family history of OA, and occupational risk. This study simplified the model using predictors only available in adolescence: age, BMI, and knee injury.
The simplified model was studied for discriminative ability using the area under receiver operating characteristics curve (AUC) and was comparable with the external validation sample of the Nottingham model. The power of the modified predication model to discriminate between men affected by knee OA and those not affected was moderate (AUC, 0.60; 95% CI, 0. 59-0.61), and the proportion of correctly classified risk was 61.9%. The study findings imply that the 40-year risk prediction for knee OA of an 18-year-old man with a BMI of 30 and a knee injury is 22% and the risk prediction for an 18-year-old man with a BMI of 25 and no previous knee injury is 7%.
A limitation of the study included OA diagnoses only from inpatient or outpatient specialist care, meaning less severe cases may have been excluded. In addition, diagnostic codes in this setting may appear differently than in primary care. A second limitation was that only data from men were used in the study, so the discriminative ability of the model may not be similar for women.
This study validated a simple prediction model for knee OA using only 3 predictors (age, BMI, and knee injury) with moderate discrimination comparable with other prediction studies that had shorter follow-up periods. A 40-year risk prediction for knee OA and the simplicity of the model may increase the tool’s utility for preventing knee OA at the individual and population level.
Reference
Magnusson K, Turkiewicz A, Timpka S, Englund M. A prediction model for the 40-year risk of knee osteoarthritis in adolescent men [published online June 28, 2018]. Arthritis Care Res. doi: 10.1002/acr.23685